Research on Development Characteristics and Landslide Dam Hazard Prediction of Zhuangfang Landslide in the Upper Reaches of the Nu River

Abstract

The upper reaches of the Nu River have strong tectonic activities and broken rock mass structures, often causing landslide disasters. The Zhuangfang landslide has apparent signs of surface deformation, and there is a risk of further sliding and blocking of the river. Taking the Zhuangfang landslide as an example, this paper analyzes the development characteristics and stability through geological field surveys, a drone aerial survey, field drilling, and GEO5 geotechnical engineering software. Then through the indoor tests and RAMMS numerical simulation software, the parameters of the landslide are determined, and the risk of a landslide dam is analyzed. Our results demonstrated that the Zhuangfang landslide is a large-scale landslide with a volume of about 4.5 × 10⁶ m³. The front edge of the landslide is seriously deformed and is in an under-stable state, with risks of sliding and river blockage. The numerical simulation results showed that the total movement time of the landslide was 130 s, and the landslide entered the Nu River at 55 s. However, the landslide does not completely block the river and cannot form a landslide dam. The study proposed a parameter inversion method to determine the landslide mass parameters based on RAMMS numerical simulation software. The related results of this study can provide a reference for the sustainable development of the ecological environment in the Nu River Basin.

1. Introduction

Landslide-induced river dam events are widely experienced in mountainous areas of the world. The disasters they bring often submerge the upstream, block the downstream, and cause secondary disasters such as floods and surges. At the same time, wild fish resources in the basin are rapidly decreased, terrestrial plants rot and die due to water immersion, and upstream and downstream villages, farmland, livestock, and infrastructure are damaged. In addition, the landslide blocking the river will form barrier lakes that silt up river sand and change river courses. After the dam breaks, it will also cause debris flow and water pollution, which pose a serious threat to the sustainable development of the ecological environment in the basin. In modern times, there have been many severe landslide dam events in the world, such as the Usoi landslide dam in Tajikistan in 1911 [1], the Val Pola landslide dam in the Italian Alps in 1987 [2], the Tsatichhu landslide dam in Bhutan in 2003 [3], and the Tangjiashan landslide dam in 2008 [4]. Notably, these landslide dams are distributed in various countries and cause serious harm to the local people.

In China, a serious landslide incident occurred recently, which was the Jinsha River Baige landslide incident in 2018. This incident affected a total of 102,000 people in the Tibet Autonomous Region, Sichuan Province, and Yunnan Province, and 86,000 people were urgently relocated. The direct economic loss of Yunnan Province alone was CNY 7.43 billion [5]. Thus, reasonable prediction of landslide dams is the key to solving this problem (Figure 1).

At present, most studies on landslide-induced river blocking focus on the failure mechanism and stability analysis of landslide dams [6,7,8]. Still, there are few studies on predicting landslide dam formation [9,10,11,12]. In recent years, many scholars have extracted geomorphic features such as material properties, valley topography, stream power conditions, or dynamic characteristics of landslides, such as landslide start-up speed, landslide mass volume, and vertical movement distance, to establish an empirical prediction model that predicts the risk of landslide dams [13,14]. However, the prediction results are often qualitative conclusions, and a certain landslide’s specific evolution range after instability cannot be drawn. At the same time, the landslides are affected by many micro-geomorphological characteristics during river blocking. Thus, the prediction results often have certain deviations [15,16,17,18].

In addition, many scholars have studied the evolution process of landslides blocking rivers through indoor experiments and established prediction formulas for landslide dams based on the experimental results [19,20]. However, the indoor test is a generalization of the actual landslide geomorphic environment, and the real geomorphic environment is far more complicated than the test device. As such, the test results are open to question [21,22].

As a main research method, the numerical simulation method can reproduce the process of a landslide blocking a river. By simulating the movement process of the landslide mass under the complex three-dimensional terrain, the specific evolution range of the landslide mass can be determined, which provides a basis for the prevention and control of landslide dam disasters [23]. The current numerical simulation research primarily focuses on a particular link in the evolution process of landslide blockage [24,25,26], but lacks the prediction research on field landslides. Moreover, it is often difficult to determine the parameters used in landslide movement.

In southwestern China, due to the long-term compression and collision of the Pacific plate and the Indian Ocean plate, tectonic faults are common. At the same time, the water system in this area is being developed. The Nu River, the Lancang River, and the Jinsha River start from the Qinghai-Tibet Plateau and flow southward through Yunnan Province [27]. Additionally, the strong erosion of tectonic movement and the downcutting action of the river lead to severe rock and soil fragmentation along both sides of the valley [28,29,30,31], which provide excellent geological conditions for the occurrence of landslide dam disasters. This paper takes the Zhuangfang landslide in Yunnan Province, China, as an example and analyzes the deformation and failure characteristics and instability scale of the Zhuangfang landslide through an on-site geological survey, drone aerial survey, and on-site drilling. The landslide stability is evaluated using GEO5 (2022.54) software. At the same time, a parameter inversion method is proposed, which determines the landslide mass parameters based on the Voellmy-fluid friction model through comparative tests of the indoor chute and numerical simulation chute. The process of the landslide blocking the river is dynamically evolved using the RAMMS (DEBRIS FLOW 1.8.0) numerical simulation software, and the risk of blocking the river is evaluated. Therefore, this study can provide a reference basis for the assessment of landslide dam disasters in the Nu River Basin.

2. Materials and Methods

2.1. Overview of the Zhuangfang Landslide

2.1.1. Background of the Landslide Area

The Zhuangfang landslide is located on Zhuangfang Road, Lushui City, Nu Prefecture, Yunnan Province, on the right bank of the Nu River. The Provincial Highway S288 and Baolu Expressway G5613 pass through the foot of the landslide area. The geographical coordinates of the center point of the landslide trailing edge are 98°83′55″ E, 25°86′46″ N (Figure 2).

The landslide area belongs to the topography of high mountains and wide valleys, and the valley’s shape is U. The structural unit belongs to the Fugong–Zhenkang fold belt of the Gangdise–Nianqing Tanggula fold system. Affected by the Himalayan movement, the crust was strongly uplifted, accelerating the undercutting of the river, resulting in steep topography, folds, and faults on both sides of the river valley. At the same time, under the action of multi-stage tectonic movement, the rock mass on both sides of the bank was broken, and a large number of loose deposits were produced, becoming the main component of the landslide mass (Figure 3).

The rainy season in the landslide area is mostly from June to October every year with abundant rainfall. The average annual rainfall is about 1200 mm, the maximum rainfall is 1742.1 mm, and the maximum daily rainfall is 105.3 mm. Additionally, there are irrigation water projects and sprinkler irrigation equipment in the landslide area to irrigate the slope farmland all year round. Rainfall and irrigation reduce the mechanical properties of the soil in the landslide area and reduce the potential sliding surface friction of the landslide. At the same time, due to the large amount of cultivated land at the front edge of the landslide, irrigation causes the soil at the front edge to continuously soften, aggravating the deformation of the front edge of the landslide area and becoming the main reason for the occurrence of landslides.

2.1.2. Landslide Details

The overall plane shape of the Zhuangfang landslide is armchair-shaped. The length of the landslide is 620 m, the transverse width is 520 m, the area is about 3 $\times$ 10⁵ m², the thickness of the landslide is 10~25 m, the average thickness is 15 m, and the total volume of the landslide is about 4.5 $\times$ 10⁶ m³, considered to be a large landslide. The overall slope of the landslide is 30°, and the slope of the trailing edge is steep, up to 50° in some areas. The elevation of the trailing edge is 1150~1140 m, the slope of the front edge is relatively gentle, with an average slope of 20°, the elevation is 920~900 m, and the maximum height difference between the front and rear edges is 250 m. The overall slope of the landslide faces northeast, and the main sliding direction is 70°. On the north side of the landslide, there is a gully with a width of 10~15 m, a depth of 5~10 m, and an extension length of 800 m. The runoff in the gully is strong, and fissures have developed. Moreover, the slope on the south side of the landslide forms a staggered platform with a height of more than 10 m. The landslide terrain at the staggered platform is steeper with an average slope of 45°.

There are apparent platform ridges between the mountains around the landslide and the landslide mass, and the shape is an inverted bell. The thick accumulation layer at the front edge of the landslide is tongue-shaped and protruding, which has an obvious diversion effect on the river. At the same time, there are a large number of isolated rocks on the landslide slope. Based on the above characteristics, it is comprehensively judged that the Zhuangfang landslide was formed by the local resurrection of an ancient landslide accumulation mass (Figure 4).

2.1.3. Stratum Structure

According to the results of on-site drilling, the strata exposed in the landslide area from new to old are Quaternary accumulative strata, middle Triassic strata, and Cambrian strata (Figure 5a,b). The landslide mass is mainly composed of the Quaternary accumulation layer (Q₄^del), and the gravel-containing silty clay layer is the most common, widely distributed layer on the landslide surface. Moreover, the occurrence of outcropping rock formations at the trailing edge is 280 ∠ 26° (Figure 5c).

There are no obvious interlayer dislocations and scratches found in the landslide strata. However, the soil structure near the interface between the landslide mass and bedrock is loose, the water content is high, and the interface between the landslide and bedrock is clear. It is speculated that the sliding surface is mainly the interface between the overlying accumulation and the bedrock, and the slope is 30°, consistent with the slope of landslide topography (Figure 5c).

The sliding bed is the underlying bedrock, mainly composed of slate (Є₃b) of the Upper Cambrian Baoshan Formation and dolomite (T₂h¹) of the Hewan Formation of the Middle Triassic System. The upper Cambrian Baoshan Formation is slate, gray, or grayish blue, and the rock formation is 268° ∠ 53°. Furthermore, the dolomite in the Hewan Formation of the Middle Triassic system is gray and off-white, and the rock formation is 261° ∠ 55° (Figure 5c).

Based on the analysis of the current stratum structure in the landslide area, the lithology of the overlying stratum of the landslide mass is silty clay with crushed stones, the structure is loose, and it easily expands and deforms when exposed to water, resulting in surface cracks. At the same time, because the permeability of the overlying soil layer is greater than that of the bedrock, the water content on the potential sliding surface is further increased, and the sliding surface is softened and damaged, which is not conducive to the stability of the landslide.

2.1.4. Deformation Characteristics

According to the on-site geological survey, the signs of deformation of the Zhuangfang landslide are mainly concentrated at the front edge and the southern boundary of the landslide. According to residents, the cracks in houses deepened during the rainy season from June to October 2019, and the width of cracks in many places reached more than 2 cm. The most severe crack was 8 cm wide and extended 2 cm. At present, it is uninhabited. The longest extension length of the pavement cracks at the front edge was 3~4 m, and the sinking distance was 1~2 cm. Additionally, the pavement cracks were densely distributed at the front edge. The retaining wall was directly sheared at the south boundary of the landslide, with a width of 70 cm and a distance of 50 cm. At the same time, a small-scale farmland collapse occurred, and the pavement on the south side also cracked. The crack was 5~6 cm wide and 7~8 m long, accompanied by large-scale subsidence of the road surface, indicating that creeping slip deformation occurred at the landslide’s front edge. Furthermore, there are a few fissures at the trailing edge of the landslide, mainly manifested as road cracks, with a width of 4 cm and an extension length of 10 m (Figure 6).

2.1.5. Stability Analysis

To further analyze the stability of the landslide, using the soil slope stability analysis module in GEO5 geotechnical engineering software, section line I-I’ was selected as the calculation section, and the stability coefficient of the landslide was calculated. Due to the concentrated and abundant rainfall and irrigation in the Zhuangfang landslide area, the front edge of the landslide area was severely deformed. Thus, calculation conditions were divided into natural and rainfall conditions. The calculation method adopted the Morgenstern–Price method of the limit equilibrium method and used the calculation results of the Spencer method and the Janbu method as verification. The landslide calculation model is shown in Figure 7. The geotechnical parameters of the landslide were determined by laboratory tests after field sampling, as shown in

Table 1. Stability calculation parameters of the Zhuangfang landslide.

Stratum	Volumetric Weight γ/(kN·m3)		Internal Friction Angle φ/(°)		Cohesion Force c/kPa
	Natural	Saturation	Natural	Saturation	Natural	Saturation
Landslide mass	21	22	36	35	18	15
Slip soil	21	22	22	20	15	14
Bedrock	25	-	41	-	700	-

.

According to the calculation results of the stability coefficient of the GEO5 geotechnical engineering software, the stability of the landslide was evaluated following the provisions of the landslide stability state in the “Code for Engineering Investigation of Landslide Prevention and Control” (GB/T32864-2016) (

Table 2. Landslide stability status classification.

Landslide Stability Coefficient (FS)	FS < 1.00	1.00 ≤ FS < 1.05	1.05 ≤ FS < 1.15	FS ≥ 1.15
Landslide stable state	unstable	Under-stable	Basically stable	stable

) [32]. The calculation results of the landslide stability coefficient are shown in

Table 3. Calculation results of the Zhuangfang landslide stability.

Calculation Conditions	Morgenstern–Price Method		Spencer Method		Janbu Method
	Stability Coefficient	Stable State	Stability Coefficient	Stable State	Stability Coefficient	Stable State
Natural	1.12	Basically stable	1.13	Basically stable	1.12	Basically stable
Rainfall	1.04	Under-stable	1.04	Under-stable	1.03	Under-stable

.

According to the calculation results in Table 3, the stability coefficient of the landslide under rainfall conditions was 1.03, and the landslide was in an under-stable state. Since the front edge of the landslide demonstrates severe deformation, if it encounters adverse geological effects such as long-term irrigation and rainfall, the landslide may be unstable, resulting in overall sliding. At the same time, the front edge of the landslide is close to the Nu River, and the shortest horizontal distance from the Nu River is only 280 m, which easily causes geological disasters in the landslide dam. Therefore, it is necessary to analyze and predict the dynamic evolution range and the risk of river blockage of the landslide after sliding.

2.2. Prediction and Analysis of Landside Dam Hazard of the Zhuangfang Landslide

2.2.1. Computational Model

The constitutive model used by the RAMMS software is the Voellmy-fluid friction model, dividing the friction resistance into two parts. One part is the dry Coulomb friction coefficient proportional to the normal stress $\mu$, and the other part is the turbulence coefficient $\xi$. The following formula expresses the friction resistance S(Pa):

$$S=\mu N+\frac{\rho g{u}^2}{\xi }+\left(1-\mu \right)N_0-(1-\mu )N_0{e}^{-\frac{N}{N_0}}$$

(1)

$$N=\rho hgcos\left(\phi \right)$$

(2)

where $N$ is the normal stress on the running surface. $\rho$ is the material density of the landslide mass, $g$ is the acceleration of gravity, $\phi$ is the slope angle, $h$ is the flow height, and $u$ is the vector velocity, including the velocity of the fluid in $X$ and $Y$ directions, $u={(u_x,u_y)}^T$. Additionally, $N_0$ is the yield stress of the material, and the formula ensures that when the normal stress $N$ and vector velocity $U$ tend to 0, the frictional resistance $S$ also tends to 0 [33].

2.2.2. Model Establishment

Establishing the numerical model in the RAMMS software is mainly realized through the digital elevation model (DEM). The accuracy of the simulation results depends on the accuracy of the input terrain data. In this study, the aerial survey of the Zhuangfang landslide area was carried out by the DJ-Innovations (DJI) Unmanned Aerial Vehicle, the route was planned by Pix4Dcapture (V4.7.0) software, and the high-precision DEM and orthophoto map of this area were generated by Pix4Dmapper software (Version 4. 4. 12). Since the DEM generated by aerial surveys is often affected by ground objects such as houses, vegetation, and roads that have nothing to do with the terrain, the Globalmapper (v22.0) software filtered the point cloud of the DEM. Moreover, the point cloud elevation near the Nu River was modified to generate a DEM that can reflect the real topography of the Zhuangfang landslide area (Figure 8).

Finally, we imported the filtered point cloud DEM, orthophoto map, and corresponding world files into the working folder of the RAMMS software and generated the final 3D geological grid model (Figure 9).

2.2.3. Parameter Calibration

In the Voellmy-fluid friction model used by the RAMMS software, friction coefficient $\mu$ and turbulence coefficient $\xi$ are two important parameters affecting landslide movement. However, these two parameters are often difficult to obtain through experiments, and current research mainly relies on inverse calculation. The RAMMS software user manual summarizes the cases of landslides and debris flows around the world and provides the recommended value range of $\mu$ and $\xi$, with $\mu$ between 0.05 and 0.4 and $\xi$ between 100 and 200 m/s².

Based on the above analysis, the authors designed a set of comparative experiments between the indoor chute and the numerical simulation chute (Figure 10). By comparing the differences in the accumulation height, accumulation distance, and retention length between the landslide and the indoor chute test, the friction coefficient $\mu$ and turbulence coefficient $\xi$ of the landslide were determined. The geometric dimensions of the indoor chute and the numerical model were consistent. Additionally, the slope of the chute was 40°, and the dimensions of the chute were 2.3 m, 0.5 m wide, and 0.5 m high. Futhermore, the accumulation area was 2 m long and 1.5 m wide, and the landslide volume was set to 0.03 m³ (Figure 11).

In the indoor chute test, the undisturbed soil samples of the Zhuangfang landslide were used for the landslide mass. Larger stones and impurities were screened out to eliminate the influence of particle size inhomogeneity on the landslide mass’s movement (Figure 10a). During the test, the baffle blocked the landslide soil. Then the baffle was rotated to release the soil. After the movement of the soil stopped in the accumulation area, the test data were recorded.

The software dump step time was set to 0.1 s. According to the results of the indoor chute test, the total time of the landslide from start to stop was 3 s. Thus, the end time of the landslide was set as 3 s in the numerical simulation (Figure 10b). Then, through the control variable method, we kept $\mu$ constant, changed the $\xi$ value according to $\pm$10 m/s², kept $\xi$ constant, and changed the $\mu$ value according to $\pm$0.01. Until a parameter value consistent with the indoor chute test results was obtained, it was used as the parameter of this numerical simulation.

Figure 12 shows the motion eigenvalues of the landslide under turbulence coefficients 100, 150, 200, and 250 m/s² and friction coefficients 0.05, 0.1, 0.15, 0.2, 0.25, 0.3, 0.35, 0.4 in the numerical simulation chute test. It can be seen from Figure 12 that the horizontal distance of the landslide accumulation is inversely proportional to the maximum accumulation height and retention length of the landslide accumulation. With the increase in friction coefficient $\mu$, the accumulation height and retention length of the landslide deposit gradually increased, and the horizontal distance gradually decreased. With the increase in turbulence coefficient $\xi$, the accumulation height and retention length of the landslide accumulation gradually decreased, and the horizontal distance increased gradually. That is, friction coefficient $\mu$ plays a controlling role in the stopping of landslide movement. Furthermore, turbulence coefficient $\xi$ plays a controlling role in the process of landslide movement.

According to the indoor chute test results, the overall shape of the landslide accumulation in the accumulation area was fan-shaped. The fan-shaped central axis was 0.6 m long, the horizontal axis was 0.75 m long, the highest accumulation height was 0.145 m (Figure 13), and the retention length in the chute was 0.27 m. Comparing the test results of the numerical simulation chute, when the friction coefficient $\mu$ was 0.33 and the turbulence coefficient $\xi$ was 150 m/s², the fan-shaped central axis was 0.58 m long (Figure 13a), the horizontal axis was 1.25 m long, the retention length in the chute was 0.26 m (Figure 13b), and the fan-shaped accumulation height was up to 0.115 m (Figure 13c). Notably, the values were basically consistent with the results of the indoor chute test. Furthermore, the distance of the horizontal axis was higher than the actual measurement result because the minimum width of the grid unit in the horizontal direction was 0.25 m.

Combined with the above analysis, the calculation parameters in the RAMMS software were finally determined as shown in

Table 4. RAMMS software calculation parameters.

Calculation Parameters	Value
DEM resolution	1 m
Release zone depth	15 m
Release zone volume	4,507,214 m3
Friction coefficient $\mu$	0.33
Turbulence coefficient $\xi$	150 m/s2
Density	2000 kg/m3
Simulation end time	1000 s
Momentum percentage	10%
Dump step time	5 s

. The stopping criterion used in the simulation calculation was the momentum percentage stopping criterion of the moving body, the release method adopted block release, and the curvature calculation was turned on.

3. Results

3.1. Landslide Movement Process

Figure 14 and Figure 15 show the height and velocity maps of the accumulation movement of the Zhuangfang landslide at different times, respectively. The total release volume of the Zhuangfang landslide was 4.5 $\times$ 10⁶ m³. The total movement time was 130 s, the maximum movement speed was 21.9 m/s, the maximum movement distance was 400 m, and the maximum accumulation height was 56.1 m. We took t = 0 s, t = 10 s, t = 30 s, t = 50 s, t = 70 s, t = 90 s, t = 110 s, and t = 130 s to discuss the specific movement process of the landslide.

From 0 to 10 s, the landslide began to slide along the sliding surface. Due to the gentle slope of the thick accumulation layer at the front edge, the accumulation height of the soil mass was the largest at this position, and the maximum accumulation height reached 32 m. The movement speed of the landslide was the largest at this stage, and the maximum movement speed was 21 m/s. The maximum velocity of the landslide occurred at the south side of the front edge of the landslide, which was due to the steep slope of the landform, with an overall slope of more than 40°. From 10 to 30 s, the maximum accumulation height of the landslide reached 36 m. Additionally, the soil mass on the south side of the landslide front began to move toward the Nu River due to the steep terrain. The shortest distance from the Nu River was 140 m in 30 s, and the maximum moving speed of the landslide at this stage was 19.5 m/s. From 30 to 50 s, the landslide gradually approached the surface of the Nu River, and the shortest distance from the Nu River was only 20 m in 50 s. The maximum accumulation height of the landslide was 37.6 m. From 50 to 90 s, part of the landslide mass began to enter the Nu River. The maximum speed when entering the Nu River was 10 m/s, and the maximum accumulation height was 15.2 m. The maximum accumulation height of the landslide was 39.3 m, and the maximum moving speed was 15.5 m/s. From 90 to 130 s, the speed of the landslide accumulation and movement slowed down continuously. The maximum accumulation height of the landslide was 39.4 m, and the maximum moving speed was 17 m/s. Furthermore, the maximum velocity of the landslide entering the Nu River was 8 m/s, the total accumulation volume was 8 $\times$ 10⁴ m³, and the maximum accumulation height was 24 m [34,35].

Figure 16 shows the characteristics of the Zhuangfang landslide accumulation in different movement areas. It can be seen that the micro-geomorphological characteristics greatly influenced the volume of the landslide mass entering the river. Most of the landslide mass was completely detained on the slope due to the influence of the terrain, accounting for 48.9% of the total landslide volume. Additionally, a small part of the landslide flowed into the Nu River, accounting for only 1.7% of the total volume of the landslide. The prediction results will often have certain deviations if the risk of the landslide dam is judged only by empirical formulas or remote sensing satellites.

3.2. Characteristics of Entering the River

Figure 17 shows the accumulation height and velocity of the Zhuangfang landslide changing with time when it enters the right bank of the Nu River. It can be seen that the Zhuangfang landslide began to enter the Nu River when it moved to 55 s, and at this time, the speed and the accumulation height of the landslide mass began to increase rapidly. At 60 s, the velocity of the landslide mass reached the maximum. The maximum velocity was 8 m/s, and the velocity then decreased gradually. In addition, the height of the landslide was continuously accumulated, and the maximum accumulation height reached 24 m at the end of the landslide movement.

Figure 18 shows the longitudinal section at the maximum accumulation height of the Nu River sliding mass. It can be seen that the channel width of the Nu River at this position was 170 m, and the altitude of the river bottom was 810 m. The maximum accumulation height of the landslide after entering the Nu River was close to the right bank of the Nu River, the maximum height was 24 m, and the altitude was 842 m. The maximum moving distance of the landslide was about 75 m. In addition, according to the calculation of the RAMMS software, the maximum longitudinal width of the landslide accumulation was about 200 m.

According to the calculation results of the RAMMS software, the shape of the accumulation after the Zhuangfang landslide enters the river is fan-shaped. and the position of entry is located on the south side of the landslide and on the right bank of the Nu River. The maximum velocity of the accumulation entering the Nu River is 8 m/s, the maximum transverse width is 200 m, the maximum longitudinal width is 75 m, the maximum accumulation height is 24 m, and the total volume of the accumulation is 0.8 $\times$ 10⁵ m³ (Figure 19).

Combined with the evolution range and movement velocity characteristics of the landslides entering the Nu River, the landslide mass did not completely block the river. Thus, the Zhuangfang landslide cannot completely form a landslide dam. However, since part of the landslide mass still enters the Nu River, the landslide may still block the river incompletely, resulting in a surge of geological disasters. In addition, the landslide impact will cause serious harm to highway G5613, provincial highway S288, about 6 km² of cultivated land at the front of the landslide, and about 50 households and 200 people in residential areas. It is recommended to carry out timely protection and control of landslides, such as setting up retaining walls or anti-slide piles in areas with severe deformation at the front edge of landslides. At the same time, since landslides are greatly affected by rainfall and irrigation, it is recommended to protect the slope surface, such as concrete slope protection and tree planting. And to strengthen landslide monitoring and early warning, devices such as landslide depth displacement detectors and earth pressure detectors can be installed.

4. Discussion

4.1. Influencing Factors of Landslide Dam Formation

The formation of a landslide dam is affected by many factors. For example, particle fragmentation during the movement of the landslide causes changes in the particle size of the landslide mass and a reduction in effective stress, which changes the movement speed and accumulation characteristics of the landslide mass [36]. After the landslide enters the river, the interaction between water and soil causes some particles to be washed away by the water flow, thus affecting the formation of the landslide dam. There are also rough and complex terrains that change the movement speed and volume of landslides, thus affecting the formation of landslide dams [37]. Although the current research on landslide dams has made great progress, few studies have been able to combine these factors; that is, the influence of factors such as water flow conditions and valley shape on the formation of landslide dams has been considered while particle breakage is considered. In addition, there are few studies that consider the influence of different factors on the formation form of landslide dams, because different dam forms may affect the stability of the dam. There are also few studies involving the formation of landslide dams and their impact on the sustainable development of the surrounding ecological environment. In this paper, numerical simulation is used to analyze the influence of the actual topography on landslide-induced river blocking, but factors such as water flow conditions and particle breakage were not considered [38]. In future research, the influence of different factors on the dam formation, dam form, and stability should be considered comprehensively [39], so as to better serve the risk prediction of landslide dams.

4.2. Parameter Calibration

Parameter calibration has always been an important problem in numerical simulation. The existing research mostly determines parameters through inversion analysis, such as comparing indoor mechanical tests with numerical simulation mechanical tests or comparing the landslide cases that have occurred with the landslide characteristics under the numerical simulation [40]. Although the current research has achieved certain results, it is still difficult to determine its parameters in the prediction of the risk of landslide dams because its accumulation characteristics and evolution range are unknown. In this paper, the friction parameters in the RAMMS numerical simulation software were determined by inversion through the comparison test of the indoor chute and the numerical simulation chute, which provided a new idea for parameter calibration in the numerical simulation [41], but its accuracy still needs to be verified by actual cases. In future research, different landslide dynamic evolution test devices can be designed for parameter inversion and compared with real landslide cases, or parameters obtained from indoor mechanical tests can be compared based on real cases to improve the accuracy of this method.

5. Conclusions

This paper comprehensively analyzed the deformation and failure characteristics and stability of the Zhuangfang landslide through a field geological survey, drone aerial survey, on-site drilling, and GEO5 geotechnical engineering software. The danger of a landslide dam was analyzed through indoor tests and numerical simulation. The following conclusions were specifically drawn:

(1)

The Zhuangfang landslide is a large landslide revived from an ancient landslide. The total volume of the landslide is about 4.5 $\times$ 10⁶ m³. The landslide is affected by rainfall and irrigation, and the front edge shows severe signs of deformation. At the same time, according to the calculation results of GEO5 software, the landslide is currently in an under-stable state, and there is a risk of further instability sliding and river blockage.

(2)

Through the comparative tests of the designed indoor chute and numerical simulation chute, the accumulation height, accumulation range, and retention length of landslide accumulation under the two groups of tests were compared by the control variable method. The landslide parameters based on the Voellmy-fluid friction model were determined, where the friction coefficient $\mu$ is 0.33 and the turbulence coefficient $\xi$ is 150 m/s². From the accumulation characteristics of the landslide, the topography significantly impacts the volume of the landslide entering the river. Most of the landslides are completely detained on the slope, and only a small part of the landslides flow into the Nu River, accounting for only 1.7% of the total volume of the landslide. From the numerical simulation results, the landslide entering the Nu River has not completely blocked the river. Thus, the Zhuangfang landslide cannot form a dam to block the river. However, because some landslides still enter the Nu River, the landslides will not completely block the river, causing river sand siltation and river diversion, leading to water pollution, fish reduction, vegetation destruction, and damage to the front disaster-bearing bodies, which is not conducive to the sustainable development of the ecological environment in the basin. It is recommended to carry out timely management of the landslide.

(3)

This paper did not consider the impact of the interaction between water and soil on the formation of landslide dams in the numerical simulation. In fact, the impact of the river has a great influence on the formation of landslide dams. How to simulate the coupling between water and soil under high-precision three-dimensional terrain is the main research direction for predicting landslide dam formation in the future. At the same time, in the numerical simulation, the determination of parameters is still a difficult problem. This paper proposes a new idea through the parameter inversion method, but its accuracy still needs more cases to be verified.